Lunar and Planetary Science XLVIII (2017)
1742.pdf
OXYGEN DEPLETION ON THE SURFACE OF MERCURY: EVIDENCE OF SILICON SMELTING?
F. M. McCubbin1, K. E. Vander Kaaden1, P. N. Peplowski2, A. S. Bell3, L. G. Evans4, L. R. Nittler5, J.W. Boyce1, L.
P. Keller1, T. J. McCoy6, 1NASA Johnson Space Center, Mailcode XI2, 2101 NASA Parkway, Houston, TX 77058,
USA. 2The Johns Hopkins University Applied Physics Laboratory, Laurel, MD 20723, USA. 3Institute of Meteoritics, Department of Earth & Planetary Sciences, University of New Mexico, Albuquerque, NM 87131, USA.
4
Computer Science Corporation, Lanham-Seabrook, MD 20706, USA. 5Department of Terrestrial Magnetism, Carnegie Institution of Washington, DC 20015, USA. 6Department of Mineral Sciences, National Museum of Natural
History, 10th and Constitution Aves. NW, Smithsonian Institution, Washington, DC 20560, USA
([email protected]).
Introduction: The MErcury Surface, Space ENvironment, GEochemistry, and Ranging (MESSENGER)
spacecraft collected data that provided important insights into the structure, chemical makeup, and compositional diversity of Mercury. The X-Ray Spectrometer
(XRS) and Gamma-Ray Spectrometer (GRS) onboard
MESSENGER provided the first detailed chemical
analyses of Mercury’s surface. Among the many discoveries included several surprising characteristics
about the surface of Mercury, including elevated S
abundances (up to 4 wt%), low Fe abundances (less
than 4 wt%), and relatively low O abundances (O/Si
ratio of 1.40±0.03) [1–3].
The surface chemistry as determined by
MESSENGER has been used to identify up to nine
distinct geochemical terranes on Mercury [4–5]. Numerous modeling and experimental efforts have been
undertaken to infer the mineralogy and petrology of
mercurian lavas and surface materials. However, all of
these efforts have presumed valence states for each of
the elements according to the following: Si4+, Ti4+,
Al3+, Cr2+, Fe2+, Mn2+, Mg2+, Ca2+, Na+, K+, S2-, Cl-.
Based on these valence assignments, cations are
charged balanced with the anions O2-, S2-, and Cl- and
the compositions are recast in terms of oxides, sulfides,
and chlorides. Based on these assumptions, the geochemical terranes that have been identified on Mercury
yield O/Si wt. ratios ranging from 1.61 to 1.84, which
is substantially higher than the preliminary O/Si ratio
of 1.40±0.03 determined by the MESSENGER GRS
[2]. We have re-evaluated the O/Si ratio using the entire MESSENGER dataset to reassess its implications
for the geochemistry of Mercury.
Results: The analysis of the O/Si ratio determined
here leverages the methodology of [2] as well as the
independent analysis techniques of [6]. We have also
revised our analysis of the 6129-keV O gamma-ray
peak to exclude contributions from the neighboring
6111-keV Cl gamma ray [7], which was not taken into
account by [2]. Our revised, northern-hemisphereaverage O/Si abundance for Mercury is 1.2±0.1. The
error represents the one-standard-deviation statistical
uncertainties in the measurement, as well as systematic
uncertainties associated with various background cor-
rections. We investigated the possibility of spatially
varying O/Si abundance ratios by binning GRS data
acquired at high-northern latitudes (e.g. primarily
northern volcanic plains material), mid-northern and
equatorial latitudes (primarily intercrater plains), and
specifically within the “high-Mg region” as defined by
MESSENGER X-Ray Spectrometer data [4]. We observed no differences for either of these spatial regions
at the one-standard-deviation level. These results confirm that the valence assignments of elements overestimate the surface abundance of O on Mercury by approximately 11.6–15.9 wt%. To make up for the O
deficit, it requires that some of the elements measured
by GRS and XRS are not bonded to O2-, S2-, or Cl-,
implying that these elements either comprise metallic
phases, carbides, or (less likely) they are bonded to
other anions like F- or N3-. Regardless, the O/Si value
that has been measured on the surface of Mercury is
either a primary (i.e., magmatic) or secondary (e.g.,
degassing or space weathering) feature, so we consider
both possibilities.
Is the O/Si ratio primary or secondary? If the
O/Si ratio is primary, we can use the geochemical behaviors of the elements in magmatic systems to determine which elemental valences have been improperly
assigned. Fe and Ti are unlikely to play much of a role
in making up for the O deficit due to their low abundances at Mercury’s surface. In fact, Si is the only major rock-forming element measured on Mercury (Si,
Mg, Al, Ca, Na) that is near the stability field of both
its oxidized (SiO2) and metallic (Si0) forms (with respect to fO2), so we have modeled the compositions of
the lithologic portions of the nine geochemical regions
assuming that there is a mixture of metallic Si and SiO2
to make up for the calculated O deficit. The resulting
compositions indicate 12.6–17.9 wt% metallic Fe-Si
and 82.1–87.4% SiO2-depleted, MgO-rich silicate melt
compositions. In fact, some of the bulk silicate compositions of the nine geochemical terranes project into the
periclase-forsterite region along an MgO-SiO2 join.
Consequently, the melting temperature required to produce such silicate compositions do not lend credence to
the possibility of the O/Si depletion being a primary
geochemical feature of mercurian lavas and surface
Lunar and Planetary Science XLVIII (2017)
1742.pdf
materials. For this reason, we do not favor a primary
origin for the O/Si ratio of Mercury’s surface.
What process has altered the O/Si ratio? By attributing the low O/Si ratio of Mercury’s surface to a
secondary process, we are limiting the process to one
that either adds metal/carbides to the surface or one
that causes the loss of oxygen from the surface. However, we will only consider O-loss processes. Of the
processes that can cause the loss of O, we consider
space weathering and magmatic degassing (from lavas
or impact melts).
Space weathering can result in the reduction of FeO
to Fe-metal. Although space weathering has been listed
as one potential mechanism to explain the O/Si depletion on Mercury [8], there is insufficient Fe in the surface materials to make up the entire deficit of O and
there is a paucity of evidence to support the formation
of Si-, Mg-, Ca-, Al-, or Na-rich metallic phases among
the many space weathering studies of planetary materials from airless bodies. Consequently, we will not consider space weathering further.
The next process we consider is loss of O from a
silicate melt by degassing; however, under reducing
conditions options are limited because the partial pressure of O2 at mercurian fO2 is exceedingly low and
H2O would be a very minor vapor species favoring H2
[9–10]. Nevertheless, recent studies have provided
evidence for abundant graphite in Mercury’s crust [11–
14], which can react with melt-species to form CO:
2Cgraphite + SiO2melt 2COgas + Si0metal
(1)
In a system with graphite and silicate melt, whether or
not this reaction should proceed to the right depends on
the partial pressure of CO. Figure 1 illustrates the pressure-dependence of CO on the fO2 of the G-CO buffer
relative to the Si-SiO2 buffer as a function of temperature. Based on these calculations and liquidus temperatures for mercurian lavas of ~1320–1360 °C [15–17],
confining pressures of more than 100 millibars of CO
are needed to prevent the formation of Si metal in the
presence of graphite. Given that the lavas are erupting
essentially into a vacuum on the surface of Mercury,
build-up of CO pressure as well as the amount of CO
production would be a function of 1) degassing efficiency, which includes the kinetics of reaction (1) and
the kinetics of bubble nucleation, and 2) the efficiency
with which the degassed CO dissipates into the vacuum. Given that Mercury is still very dark (i.e., graphite
is likely present) and still has enough O such that 55–
75% of the Si is hosted by silicates, reaction (1) did not
proceed to completion. However, as discussed above,
conversion of SiO2 to Si metal would increase the
Mg/Si ratio of the silicate melt (Mg/Si ratio of the melt
increased by a factor of 1.5-2.2), which in turn, would
increase the temperature of the solidus, resulting in
crystallization and possibly quenching of the lavas.
Figure 1. Plot of oxygen fugacity (fO2) vs temperature
(T). Dashed lines represent the fO2 of the GCO buffer at
specified partial pressures of CO. The solid black line represents the fO2 of the Si-SiO2 buffer. Thermodynamic data for
the calculations obtained from JANAF tables [18]
Conclusion: The low O/Si ratio measured by GRS
is likely the result of a secondary process, possibly
related to magmatic degassing of CO in a process similar to industrial smelting. Although natural smelting
processes have occurred on Earth (e.g., Disko Island,
Greenland) and on asteroidal parent bodies (e.g.,
ureilite parent body), this process is unique on Mercury
because, instead of Fe, it produced metals of typically
lithophile elements like Si. This process is the natural
consequence of magmatic processes on an airless body
with graphite-bearing, low-FeO lavas. Mercury continues to demonstrate ways in which it represents a geochemical endmember in our solar system.
References: [1] Nittler et al. (2011) Science 333,
1847–1850. [2] Evans et al. (2012) JGR-Planets 117,
E00L07. [3] Weider et al. (2014) Icarus 235, 170–186.
[4] Weider et al. (2015) EPSL 416, 109–120. [5]
Vander Kaaden et al. (2017) Icarus, in press. [6]
Peplowski et al. (2015) MAPS 50, 353–367. [7] Evans
et al. (2015) Icarus 257, 417–427. [8] McCoy et al (in
revision) in Mercury as Revealed by MESSENGER,
editors S. Solomon, L. Nittler L., and B. Anderson,
Cambridge University Press. [9] McCubbin et al.
(2012) GRL 39, L09202. [10] Zolotov et al. (2013)
JGR-Planets 118, 1–9. [11] Peplowski et al. (2015)
Planetary and Space Science 108, 98–107. [12]
Murchie et al (2015) Icarus 254, 287–305. [13] Vander
Kaaden and McCubbin (2015) JGR-Planets 120, 1–15.
[14] Peplowski et al. (2016) Nature Geoscience 9,
273–276. [15] Vander Kaaden and McCubbin (2016)
GCA 173, 246–263. [16] Sehlke and Whittington
(2016) JGR-Planets 120, 1924–1955. [17] Namur et
al. (2016) EPSL 439, 117–128. [18] Chase (1998) J.
Phys. Chem. Reference Data, Monograph 9.